Coherentist Justification in Epistemology: A Shift from Foundationalism

How do we know that what we believe is actually justified? This is one of the oldest and most contested questions in epistemology – the branch of philosophy concerned with the nature and scope of knowledge. For centuries, foundationalism dominated the answer: knowledge rests on a bedrock of basic, self-evident beliefs from which everything else is derived. But a growing number of philosophers came to question whether such a bedrock even exists. The result was a significant pivot toward coherentism – a view that reimagines justification not as a pyramid built on solid ground, but as a web of mutually supporting beliefs. This post explores that shift, what it means for knowledge, and where it leaves us on the question of relativism.

The foundationalist picture and its problems 🔗

To appreciate coherentism, it helps to understand what it was reacting against. Foundationalism holds that our knowledge has a hierarchical structure. At the base sit a small set of basic beliefs – propositions that are self-evident, indubitable, or directly given through sensory experience. Every other belief is justified by tracing it back, step by step, to these foundations. The classic exemplar is René Descartes, whose famous cogito – “I think, therefore I am” – was meant to serve as an unshakeable starting point from which all other knowledge could be reconstructed.

The appeal of this model is intuitive: if beliefs need to be justified, and justification requires reasons, then the chain of reasons must stop somewhere – at beliefs that do not themselves require further justification. According to the Stanford Encyclopedia of Philosophy, a foundationalist tradition inspired by Descartes holds that justified beliefs are those that are either self-evidently true or deduced from self-evident truths. The difficulty is that very little of what we ordinarily take ourselves to know actually meets this demanding standard.

Empiricist versions of foundationalism – those that ground basic beliefs in raw sense data – face a different but equally sharp problem: sensory data, when understood as purely pre-conceptual experience, turns out to be either not informative enough to justify substantive beliefs, or not genuinely indubitable once conceptual interpretation enters the picture. Both the rationalist and empiricist branches of foundationalism thus run into trouble, and this double failure opened the door for an alternative.

The coherentist alternative: a web of belief 🔗

As articulated in the Stanford Encyclopedia of Philosophy, many epistemologists concluded that there must be something fundamentally wrong with framing justification in foundationalist terms. Rather than modeling knowledge on Euclidean geometry – with its axiomatic base and derived theorems – these thinkers favored a holistic picture in which no beliefs are privileged as foundational. All beliefs, on this view, are equal members of a web of belief.

This is the core claim of coherentism: a belief is epistemically justified by cohering with other beliefs. Coherence here is not just logical consistency – it also involves explanatory relations, where beliefs mutually support and make sense of one another, and various inductive connections. As explained in the Internet Encyclopedia of Philosophy, coherentism typically involves three components: logical consistency, explanatory relations, and various inductive (non-explanatory) relations. The stronger the web of interconnected beliefs, the more any individual belief within it is justified.

Crucially, coherentism does not ask whether a belief can be traced back to some unquestionable foundation. Instead, it asks: does this belief fit – does it cohere – with the broader system of beliefs the person holds? A belief that contradicts many other well-supported beliefs is poorly justified; one that integrates smoothly and reinforces others is better justified. Justification, in other words, is a systemic property, not an intrinsic one.

The regress problem and coherentism’s answer 🔗

Much of the motivation for coherentism comes from its response to the classic regress problem. The problem runs as follows: if every belief requires justification from another belief, and that belief requires further justification, we seem to face an infinite regress – an endless chain of “but why do you believe that?” Foundationalists and coherentists agree that the regress can be avoided, but disagree about how. Foundationalists terminate the regress by positing basic beliefs that are self-justifying. Coherentists deny that any such stopping point is needed – or even possible. Instead, they argue, what makes a belief system justified is its overall coherence. As epistemologist Paul Moser notes, coherentism affirms that all justified beliefs are justified in virtue of their relations to other beliefs – none are justified in isolation or independently.

Neurath’s boat: knowledge as ongoing repair 🔗

No image captures the coherentist vision more compellingly than Neurath’s boat – a metaphor developed by the philosopher of science Otto Neurath. Writing in 1921, Neurath described the human situation with respect to knowledge as follows: we are like sailors who must rebuild their ship while at sea, unable to start over in dry dock. Every plank that needs replacing must be replaced while the rest of the ship holds us afloat. We can never strip the vessel down to nothing and begin from scratch.

The metaphor is a direct challenge to foundationalism. It denies the possibility – or the necessity – of ever reaching bedrock. Neurath’s boat was later popularized by the philosopher W.V. Quine, who extended it to his broader naturalistic epistemology. Quine proposed the doctrine of confirmation holism, according to which the entire body of empirical knowledge is tested with every new observation, and no part of it is immune to revision – even logic can be revised if that is required for the stability of the system as a whole.

Applied to coherentist justification, Neurath’s boat conveys that we are always already in the middle of a belief system. We do not and cannot begin knowledge from zero. When new evidence appears – an anomaly, a contradiction, a discovery – we revise parts of our belief system while keeping the rest intact. The revision is local but the justificatory context is always global. No single belief is immune from revision, but no belief is revised in isolation either. The ship stays afloat precisely because we repair it plank by plank, not all at once.

The isolation objection and coherentism’s main vulnerability 🔗

Critics of coherentism have raised a pointed objection: if justification is purely a matter of internal coherence, what connects a belief system to reality at all? A richly coherent set of beliefs could be entirely fictional – a well-written novel is internally coherent, yet no one thinks it constitutes knowledge. This is the isolation objection. As the Stanford Encyclopedia notes, coherentism seems to allow coherence to imply justification even when a belief system is completely cut off from an individual’s direct experience of the world.

The related alternative systems objection sharpens this worry: for any coherent belief system, one can conceivably construct another equally coherent but incompatible system. If coherence alone suffices for justification, then both systems would be equally justified – which seems to undermine the very idea that justification tracks truth.

Most coherentists respond by granting a limited role to experience. Beliefs that arise spontaneously from perception – what Laurence BonJour called “cognitively spontaneous beliefs” – are given some initial, pre-coherentist weight. The coherentist framework then amplifies or diminishes this initial weight depending on how well these beliefs cohere with the broader system. This leads some theorists to classify such positions as versions of weak foundationalism – since some beliefs are granted a non-inferential, experience-based starting point, even if full justification remains holistic.

Does coherentism lead to epistemic relativism? 🔗

One of the most serious implications critics raise against coherentism is the possibility of epistemic relativism. If knowledge is justified by the internal coherence of a belief system, and different individuals or communities can hold different but internally coherent systems, then it seems to follow that all such systems are equally justified. There would be no neutral standpoint from which to adjudicate between them – and the concept of objective knowledge would lose its grip.

This worry is not entirely unfounded. Coherentism does imply that the same belief could be justified for one person and unjustified for another, depending on the other beliefs each holds. But most coherentists resist the full slide into relativism. They argue that not all coherent systems are equally good – some are richer, more comprehensive, more consistent with experience, and more explanatorily powerful than others. The coherence relation provides constraints on what counts as justified belief; it is not a free pass for any internally consistent worldview.

As explored in a study published in Studies in Philosophy and Education, the concern about relativism is particularly acute in contexts where public discourse is already fraught with misinformation. Coherentist epistemology must therefore grapple seriously with how to maintain that some belief systems are better justified than others, even without appealing to a privileged foundation.

Toward a moderate position: foundherentism 🔗

The tension between foundationalism and coherentism has prompted several philosophers to search for a middle ground. The most developed such position is foundherentism, a term coined by philosopher Susan Haack in her 1993 work Evidence and Inquiry. Foundherentism combines elements from both rival theories while attempting to address their respective problems – foundationalism’s susceptibility to an arbitrary stopping point and coherentism’s susceptibility to circularity.

Haack’s key insight is that the two theories need not be mutually exclusive. It is possible to include the relevance of experience for the justification of empirical beliefs – as foundationalism does but coherentism does not – while also allowing for pervasive mutual dependence among beliefs, as coherentism does but foundationalism does not. To illustrate this, Haack introduces the analogy of a crossword puzzle: entries provide mutual support for one another (coherence), yet some entries are also partially constrained by the clues given (a foundational element), so there is no vicious circularity.

In foundherentism, some beliefs – particularly those grounded in perception, memory, and introspection – carry more initial evidential weight than others. But these beliefs are not immune from revision, and their ultimate justificatory status depends on how well they integrate with the broader belief system. This is an important nuance: beliefs can be more foundational without being incorrigible foundations. The difference matters because it preserves a degree of epistemic hierarchy without surrendering the holistic insight that no belief is ever fully justified in isolation.

What foundherentism preserves from each side 🔗

From foundationalism, foundherentism preserves the idea that experience plays an irreplaceable role in grounding empirical knowledge. From coherentism, it preserves the idea that beliefs are always evaluated against a background of other beliefs – that justification is never a purely one-directional affair. The result is a more nuanced account that avoids both the arbitrariness of pure foundationalism and the circularity concerns that plague strong coherentism.

Foundherentism is not without its critics. Some philosophers, such as Peter Tramel writing in Synthese, have argued that foundherentism is, despite Haack’s claims, ultimately a version of foundationalism – because it still assigns a privileged role to experiential states. Others question whether the crossword analogy actually escapes the regress problem or simply relocates it. Still, foundherentism represents one of the most rigorous attempts to move epistemology beyond the binary of either pure foundations or pure coherence.

What this shift means for epistemology 🔗

The movement from foundationalism to coherentism – and toward hybrid positions like foundherentism – reflects a broader maturation in epistemology. The Cartesian dream of building knowledge from a small set of unshakeable truths has given way to a more pragmatic and dynamic picture. Knowledge is not a building with a fixed foundation; it is more like Neurath’s boat – always at sea, always being repaired, never starting from scratch.

This shift has real consequences. It changes what it means to revise a belief: revision is not a matter of discarding a faulty brick and replacing it, but of reworking part of a living structure while attending to its overall integrity. It also changes what it means to challenge someone’s knowledge claim: rather than asking “what is your basic belief here?”, the coherentist asks “how does this belief sit with everything else you hold?” The burden of justification is distributed, not concentrated at a single point.

As the Internet Encyclopedia of Philosophy notes, coherentism likely enjoys its widest acceptance in moral epistemology, particularly through John Rawls’s model of wide reflective equilibrium – a method of justifying moral principles by seeking coherence between our considered moral judgments, moral principles, and background theories. This confirms that the coherentist approach has found purchase well beyond abstract debates about perception and logic, shaping how we think about ethical knowledge as well.

Whether one ends up a committed coherentist, a moderate foundationalist, or a foundherentist, the central lesson is clear: the justification of knowledge is a holistic affair. No belief floats free, and no belief is an island. What we know is always embedded in – and sustained by – the larger web of what we believe.

What do you think? If no belief can be fully justified in isolation – only within a coherent system – does that mean our knowledge is always provisional and never truly certain? And if different cultures can maintain distinct but internally coherent belief systems, is there any principled way to say one system is more epistemically justified than another?

References

1. https://iep.utm.edu/found-ep
2. https://plato.stanford.edu/entries/justep-coherence/
3. https://iep.utm.edu/coherentism-in-epistemology/
4. https://link.springer.com/chapter/10.1007/978-94-009-4526-5_3
5. https://en.wikipedia.org/wiki/Neurath%27s_boat
6. https://neurathsboat.blog/post/neuraths-boat/
7. https://link.springer.com/article/10.1007/s11217-023-09909-0
8. https://en.wikipedia.org/wiki/Foundherentism